Skip to content

sharibkabani/AIMLSoccerTracker

Repository files navigation

MLFootballTracker

output_video.mp4

Description:

This project demonstrates how to build a comprehensive football analysis system using state-of-the-art AI and ML techniques.

The system is capable of:

  • Detecting and tracking players, referees, and the football throughout an entire video using the YOLO object detection model.
  • Training the YOLO model to improve its accuracy on specific football datasets.
  • Assigning players to teams based on their shirt colors using image segmentation and clustering.
  • Calculating team ball acquisition percentage using optical flow.

Key Features:

Object Detection: Utilizes the YOLO object detection model to accurately identify and locate players, referees, and the football in video frames.

Model Training: Provides instructions for training the YOLO model on custom football datasets to enhance performance.

Player Assignment: Employs image segmentation and clustering techniques to segment player shirts and assign them to the correct teams.

Team Analysis: Calculates team ball acquisition percentage to assess offensive performance.

Software:

Python: The primary programming language used for this project.

OpenCV: A powerful open-source computer vision library for image and video processing tasks.

YOLO: A state-of-the-art object detection algorithm implemented in TensorFlow.

scikit-learn: A machine learning library for tasks such as clustering and data analysis.

Jupyter Notebook: An interactive environment for developing and testing code.

Git: A version control system for tracking changes to the project's code.

Libraries:

NumPy: A fundamental library for numerical computing in Python.

Matplotlib: A plotting library for creating visualizations.

Pillow: A Python Imaging Library for image manipulation.

pandas: A data analysis library for working with structured data.

About

AI/ML Soccer Tracker

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published